Publicación:
Estudio, análisis y posibles mejoras de redes neuronales tipo YOLO Pose y de su uso y aplicación en la creación de una semántica común para la evaluación de pacientes con deficiencias psico-motoras

Vista sencilla de ítem
dc.contributor.advisorSantos, Olga C.
dc.contributor.authorGarcía Hernández, Victor Jesús
dc.date.accessioned2025-10-09T12:13:18Z
dc.date.available2025-10-09T12:13:18Z
dc.date.issued2024-06-10
dc.description.abstractThis work investigates the use of neural networks for key point detection to analyze movements and posture in physical therapy, focusing on the YOLO Pose algorithm. The objectives were to evaluate and improve the accuracy of YOLO Pose in identifying joint positions and to develop a semantic framework translating this data into clinically relevant parameters for assessing patients with psychomotor deficiencies. The study comprises several key stages. Initially, the accuracy of YOLO Pose was evaluated using the YOLOv8x-pose-p6 network on images representing common physical therapy movements. This analysis revealed opportunities to enhance accuracy in certain postures and conditions. Subsequently, multiple versions of the YOLOv8x-pose-p6 network were modified and trained, incorporating advanced techniques such as Batch Normalization and ReLU to improve stability and efficiency. The training utilized a dataset of labeled images showing various movements, and the best-performing configurations were identified based on accuracy. Results showed significant improvements in YOLO Pose's ability to detect key points under varied conditions, essential for clinical application. A critical aspect of this research is the development of a comprehensive semantic framework that translates raw movement data from YOLO Pose into clinically meaningful parameters for physical therapy assessment. This framework converts detected key points into metrics like ranges of motion, speed of joint movements, and gait patterns, which are crucial for clinical evaluation. The semantic framework could facilitate the integration of computer vision data into physical therapy by mapping joint positions and movements into specific ranges of motion, allowing precise measurement of flexibility and mobility. It also calculates the speed and acceleration of joint movements, providing insights into motor control and coordination. Furthermore, the framework supports the standardization of assessments across different therapists and clinical settings, ensuring consistent and comparable evaluations.en
dc.identifier.citationGarcía Hernández, Victor Jesús. Trabajo Fin de Máster: Estudio, análisis y posibles mejoras de redes neuronales tipo YOLO Pose y de su uso y aplicación en la creación de una semántica común para la evaluación de pacientes con deficiencias psico-motoras. Universidad Nacional de Educación a Distancia (UNED), 2024
dc.identifier.urihttps://hdl.handle.net/20.500.14468/30359
dc.language.isoen
dc.relation.centerE.T.S. de Ingeniería Informática
dc.relation.degreeMáster universitario en Investigación en Inteligencia Artificial
dc.relation.departmentInteligencia Artificial
dc.rightsinfo:eu-repo/semantics/openAccess
dc.rights.urihttp://creativecommons.org/licenses/by-nc-nd/4.0/deed.es
dc.subject33 Ciencias Tecnológicas
dc.titleEstudio, análisis y posibles mejoras de redes neuronales tipo YOLO Pose y de su uso y aplicación en la creación de una semántica común para la evaluación de pacientes con deficiencias psico-motorases
dc.typetesis de maestríaes
dc.typemaster thesisen
dspace.entity.typePublication
Archivos
Bloque original
Mostrando 1 - 1 de 1
Miniatura
Nombre:
Garcia_Hernandez_VictorJesus_TFM.pdf
Tamaño:
2.4 MB
Formato:
Adobe Portable Document Format
Descargar
Bloque de licencias
Mostrando 1 - 1 de 1
No hay miniatura disponible
Nombre:
license.txt
Tamaño:
3.62 KB
Formato:
Item-specific license agreed to upon submission
Descripción:
Descargar
Colecciones
Trabajos de fin de máster (TFM)